Computational Techniques for Spectral Image Analysis

Multispectral images are becoming more common in the field of remote sensing, computer vision, and industrial applications. Due to the high accuracy of the multispectral information, it can be used as an important quality factor in the inspection of industrial products. Recently, the development on multispectral imaging systems and the computational analysis on the multispectral images have been the focus of a growing interest. In this thesis, three areas of multispectral image analysis are considered. First, a method for analyzing multispectral textured images was developed. The method is based on a spectral cooccurrence matrix, which contains information of the joint distribution of spectral classes in a spectral domain. Next, a procedure for estimating the illumination spectrum of the color images was developed. Proposed method can be used, for example, in color constancy, color correction, and in the content based search from color image databases. Finally, color filters for the optical pattern recognition were designed, and a prototype of a spectral vision system was constructed. The spectral vision system can be used to acquire a low dimensional component image set for the two dimensional spectral image reconstruction. The data obtained by the spectral vision system is small and therefore convenient for storing and transmitting a spectral image.

A Quantitative method for the characterization of lytic metastases of the bone from radiographic images

The aim of our study was to assess the diagnostic usefulness of the gray level parameters to distinguish osteolytic lesions using radiological images. Materials and Methods: A retrospective study was carried out. A total of 76 skeletal radiographs of osteolytic metastases and 67 radiographs of multiple myeloma were used. The cases were classified into nonflat (MM1 and OL1) and flat bones (MM2 and OL2). These radiological images were analyzed by using a computerized method. The parameters calculated were mean, standard deviation, and coefficient of variation (MGL, SDGL, and CVGL) based on gray level histogram analysis of a region-of-interest.Diagnostic utility was quantified bymeasurement of parameters on osteolyticmetastases andmultiplemyeloma, yielding quantification of area under the receiver operating characteristic (ROC) curve (AUC). Results: Flat bone groups (MM2 and OL2) showed significant differences in mean values of MGL ( = 0.048) and SDGL ( = 0.003). Their corresponding values of AUC were 0.758 for MGL and 0.883 for SDGL in flat bones. In nonflat bones these gray level parameters do not show diagnostic ability. Conclusion: The gray level parametersMGL and SDGL show a good discriminatory diagnostic ability to distinguish between multiple myeloma and lytic metastases in flat bones.

Anàlisi d'imatges estereoscòpiques utilitzant mètodes de baixa càrrega computacional

Mitjançant imatges estereoscòpiques es poden detectar la posició respecte de la càmera dels objectes que apareixen en una escena. A partir de les diferències entre les imatges captades pels dos objectius es pot determinar la profunditat dels objectes. Existeixen diversitat de tècniques de visió artificial que permeten calcular la localització dels objectes, habitualment amb l’objectiu de reconstruir l’escena en 3D. Aquestes tècniques necessiten una gran càrrega computacional, ja que utilitzen mètodes de comparació bidimensionals, i per tant, no es poden utilitzar per aplicacions en temps real. En aquest treball proposem un nou mètode d’anàlisi de les imatges estereoscòpiques que ens permeti obtenir la profunditat dels objectes d’una escena amb uns resultats acceptables. Aquest nou mètode es basa en transformar la informació bidimensional de la imatge en una informació unidimensional per tal de poder fer la comparació de les imatges amb un baix cost computacional, i dels resultats de la comparació extreure’n la profunditat dels objectes dins l’escena. Això ha de permetre, per exemple, que aquest mètode es pugui implementar en un dispositiu autònom i li permeti realitzar operacions de guiatge a través d’espais interiors i exteriors.

Inversion and assimilation methods with applications in geophysical remote sensing

Stratospheric ozone can be measured accurately using a limb scatter remote sensing technique at the UV-visible spectral region of solar light. The advantages of this technique includes a good vertical resolution and a good daytime coverage of the measurements. In addition to ozone, UV-visible limb scatter measurements contain information about NO2, NO3, OClO, BrO and aerosols. There are currently several satellite instruments continuously scanning the atmosphere and measuring the UVvisible region of the spectrum, e.g., the Optical Spectrograph and Infrared Imager System (OSIRIS) launched on the Odin satellite in February 2001, and the Scanning Imaging Absorption SpectroMeter for Atmospheric CartograpHY (SCIAMACHY) launched on Envisat in March 2002. Envisat also carries the Global Ozone Monitoring by Occultation of Stars (GOMOS) instrument, which also measures limb-scattered sunlight under bright limb occultation conditions. These conditions occur during daytime occultation measurements. The global coverage of the satellite measurements is far better than any other ozone measurement technique, but still the measurements are sparse in the spatial domain. Measurements are also repeated relatively rarely over a certain area, and the composition of the Earth’s atmosphere changes dynamically. Assimilation methods are therefore needed in order to combine the information of the measurements with the atmospheric model. In recent years, the focus of assimilation algorithm research has turned towards filtering methods. The traditional Extended Kalman filter (EKF) method takes into account not only the uncertainty of the measurements, but also the uncertainty of the evolution model of the system. However, the computational cost of full blown EKF increases rapidly as the number of the model parameters increases. Therefore the EKF method cannot be applied directly to the stratospheric ozone assimilation problem. The work in this thesis is devoted to the development of inversion methods for satellite instruments and the development of assimilation methods used with atmospheric models.

Kernel-Based Ranking. Methods for Learning and Performance Estimation

Machine learning provides tools for automated construction of predictive models in data intensive areas of engineering and science. The family of regularized kernel methods have in the recent years become one of the mainstream approaches to machine learning, due to a number of advantages the methods share. The approach provides theoretically well-founded solutions to the problems of under- and overfitting, allows learning from structured data, and has been empirically demonstrated to yield high predictive performance on a wide range of application domains. Historically, the problems of classification and regression have gained the majority of attention in the field. In this thesis we focus on another type of learning problem, that of learning to rank. In learning to rank, the aim is from a set of past observations to learn a ranking function that can order new objects according to how well they match some underlying criterion of goodness. As an important special case of the setting, we can recover the bipartite ranking problem, corresponding to maximizing the area under the ROC curve (AUC) in binary classification. Ranking applications appear in a large variety of settings, examples encountered in this thesis include document retrieval in web search, recommender systems, information extraction and automated parsing of natural language. We consider the pairwise approach to learning to rank, where ranking models are learned by minimizing the expected probability of ranking any two randomly drawn test examples incorrectly. The development of computationally efficient kernel methods, based on this approach, has in the past proven to be challenging. Moreover, it is not clear what techniques for estimating the predictive performance of learned models are the most reliable in the ranking setting, and how the techniques can be implemented efficiently. The contributions of this thesis are as follows. First, we develop RankRLS, a computationally efficient kernel method for learning to rank, that is based on minimizing a regularized pairwise least-squares loss. In addition to training methods, we introduce a variety of algorithms for tasks such as model selection, multi-output learning, and cross-validation, based on computational shortcuts from matrix algebra. Second, we improve the fastest known training method for the linear version of the RankSVM algorithm, which is one of the most well established methods for learning to rank. Third, we study the combination of the empirical kernel map and reduced set approximation, which allows the large-scale training of kernel machines using linear solvers, and propose computationally efficient solutions to cross-validation when using the approach. Next, we explore the problem of reliable cross-validation when using AUC as a performance criterion, through an extensive simulation study. We demonstrate that the proposed leave-pair-out cross-validation approach leads to more reliable performance estimation than commonly used alternative approaches. Finally, we present a case study on applying machine learning to information extraction from biomedical literature, which combines several of the approaches considered in the thesis. The thesis is divided into two parts. Part I provides the background for the research work and summarizes the most central results, Part II consists of the five original research articles that are the main contribution of this thesis.

Application of Computational Fluid Dynamics in Modeling Blood Flow in Human Thoracic Aorta

The aim of this study was to simulate blood flow in thoracic human aorta and understand the role of flow dynamics in the initialization and localization of atherosclerotic plaque in human thoracic aorta. The blood flow dynamics in idealized and realistic models of human thoracic aorta were numerically simulated in three idealized and two realistic thoracic aorta models. The idealized models of thoracic aorta were reconstructed with measurements available from literature, and the realistic models of thoracic aorta were constructed by image processing Computed Tomographic (CT) images. The CT images were made available by South Karelia Central Hospital in Lappeenranta. The reconstruction of thoracic aorta consisted of operations, such as contrast adjustment, image segmentations, and 3D surface rendering. Additional design operations were performed to make the aorta model compatible for the numerical method based computer code. The image processing and design operations were performed with specialized medical image processing software. Pulsatile pressure and velocity boundary conditions were deployed as inlet boundary conditions. The blood flow was assumed homogeneous and incompressible. The blood was assumed to be a Newtonian fluid. The simulations with idealized models of thoracic aorta were carried out with Finite Element Method based computer code, while the simulations with realistic models of thoracic aorta were carried out with Finite Volume Method based computer code. Simulations were carried out for four cardiac cycles. The distribution of flow, pressure and Wall Shear Stress (WSS) observed during the fourth cardiac cycle were extensively analyzed. The aim of carrying out the simulations with idealized model was to get an estimate of flow dynamics in a realistic aorta model. The motive behind the choice of three aorta models with distinct features was to understand the dependence of flow dynamics on aorta anatomy. Highly disturbed and nonuniform distribution of velocity and WSS was observed in aortic arch, near brachiocephalic, left common artery, and left subclavian artery. On the other hand, the WSS profiles at the roots of branches show significant differences with geometry variation of aorta and branches. The comparison of instantaneous WSS profiles revealed that the model with straight branching arteries had relatively lower WSS compared to that in the aorta model with curved branches. In addition to this, significant differences were observed in the spatial and temporal profiles of WSS, flow, and pressure. The study with idealized model was extended to study blood flow in thoracic aorta under the effects of hypertension and hypotension. One of the idealized aorta models was modified along with the boundary conditions to mimic the thoracic aorta under the effects of hypertension and hypotension. The results of simulations with realistic models extracted from CT scans demonstrated more realistic flow dynamics than that in the idealized models. During systole, the velocity in ascending aorta was skewed towards the outer wall of aortic arch. The flow develops secondary flow patterns as it moves downstream towards aortic arch. Unlike idealized models, the distribution of flow was nonplanar and heavily guided by the artery anatomy. Flow cavitation was observed in the aorta model which was imaged giving longer branches. This could not be properly observed in the model with imaging containing a shorter length for aortic branches. The flow circulation was also observed in the inner wall of the aortic arch. However, during the diastole, the flow profiles were almost flat and regular due the acceleration of flow at the inlet. The flow profiles were weakly turbulent during the flow reversal. The complex flow patterns caused a non-uniform distribution of WSS. High WSS was distributed at the junction of branches and aortic arch. Low WSS was distributed at the proximal part of the junction, while intermedium WSS was distributed in the distal part of the junction. The pulsatile nature of the inflow caused oscillating WSS at the branch entry region and inner curvature of aortic arch. Based on the WSS distribution in the realistic model, one of the aorta models was altered to induce artificial atherosclerotic plaque at the branch entry region and inner curvature of aortic arch. Atherosclerotic plaque causing 50% blockage of lumen was introduced in brachiocephalic artery, common carotid artery, left subclavian artery, and aortic arch. The aim of this part of the study was first to study the effect of stenosis on flow and WSS distribution, understand the effect of shape of atherosclerotic plaque on flow and WSS distribution, and finally to investigate the effect of lumen blockage severity on flow and WSS distributions. The results revealed that the distribution of WSS is significantly affected by plaque with mere 50% stenosis. The asymmetric shape of stenosis causes higher WSS in branching arteries than in the cases with symmetric plaque. The flow dynamics within thoracic aorta models has been extensively studied and reported here. The effects of pressure and arterial anatomy on the flow dynamic were investigated. The distribution of complex flow and WSS is correlated with the localization of atherosclerosis. With the available results we can conclude that the thoracic aorta, with complex anatomy is the most vulnerable artery for the localization and development of atherosclerosis. The flow dynamics and arterial anatomy play a role in the localization of atherosclerosis. The patient specific image based models can be used to diagnose the locations in the aorta vulnerable to the development of arterial diseases such as atherosclerosis.

Identification of croplands of winter cereals in Rio Grande do Sul state, Brazil, through unsupervised classification of normalized difference vegetation index images

This study aimed to propose methods to identify croplands cultivated with winter cereals in the northern region of Rio Grande do Sul State, Brazil. Thus, temporal profiles of Normalized Difference Vegetation Index (NDVI) from MODIS sensor, from April to December of the 2000 to 2008, were analyzed. Firstly, crop masks were elaborated by subtracting the minimum NDVI image (April to May) from the maximum NDVI image (June to October). Then, an unsupervised classification of NDVI images was carried out (Isodata), considering the crop mask areas. According to the results, crop masks allowed the identification of pixels with greatest green biomass variation. This variation might be associated or not with winter cereals areas established to grain production. The unsupervised classification generated classes in which NDVI temporal profiles were associated with water bodies, pastures, winter cereals for grain production and for soil cover. Temporal NDVI profiles of the class winter cereals for grain production were in agree with crop patterns in the region (developmental stage, management standard and sowing dates). Therefore, unsupervised classification based on crop masks allows distinguishing and monitoring winter cereal crops, which were similar in terms of morphology and phenology.

Using fraction images derived from modis data for coffee crop mapping

Coffee production was closely linked to the economic development of Brazil and, even today, coffee is an important product of the national agriculture. The State of Minas Gerais currently accounts for 52% of the whole coffee area in Brazil. Remote sensing data can provide information for monitoring and mapping of coffee crops, faster and cheaper than conventional methods. In this context, the objective of this study was to assess the effectiveness of coffee crop mapping in Monte Santo de Minas municipality, Minas Gerais State, Brazil, from fraction images derived from MODIS data, in both dry and rainy seasons. The Spectral Linear Mixing Model was used to derive fraction images of soil, coffee, and water/shade. These fraction images served as input data for the supervised automatic classification using the SVM - Support Vector Machine approach. The best results concerning Overall Accuracy and Kappa Index were obtained in the classification of the dry season, with 67% and 0.41, respectively.

Simulation methods in the modelling of bioaffinity assays

Computational model-based simulation methods were developed for the modelling of bioaffinity assays. Bioaffinity-based methods are widely used to quantify a biological substance in biological research, development and in routine clinical in vitro diagnostics. Bioaffinity assays are based on the high affinity and structural specificity between the binding biomolecules. The simulation methods developed are based on the mechanistic assay model, which relies on the chemical reaction kinetics and describes the forming of a bound component as a function of time from the initial binding interaction. The simulation methods were focused on studying the behaviour and the reliability of bioaffinity assay and the possibilities the modelling methods of binding reaction kinetics provide, such as predicting assay results even before the binding reaction has reached equilibrium. For example, a rapid quantitative result from a clinical bioaffinity assay sample can be very significant, e.g. even the smallest elevation of a heart muscle marker reveals a cardiac injury. The simulation methods were used to identify critical error factors in rapid bioaffinity assays. A new kinetic calibration method was developed to calibrate a measurement system by kinetic measurement data utilizing only one standard concentration. A nodebased method was developed to model multi-component binding reactions, which have been a challenge to traditional numerical methods. The node-method was also used to model protein adsorption as an example of nonspecific binding of biomolecules. These methods have been compared with the experimental data from practice and can be utilized in in vitro diagnostics, drug discovery and in medical imaging.

Characterization of fiber and vessel elements in pulp suspension images

The thesis is related to the topic of image-based characterization of fibers in pulp suspension during the papermaking process. Papermaking industry is focusing on process control optimization and automatization, which makes it possible to manufacture highquality products in a resource-efficient way. Being a part of the process control, pulp suspension analysis allows to predict and modify properties of the end product. This work is a part of the tree species identification task and focuses on analysis of fiber parameters in the pulp suspension at the wet stage of paper production. The existing machine vision methods for pulp characterization were investigated, and a method exploiting direction sensitive filtering, non-maximum suppression, hysteresis thresholding, tensor voting, and curve extraction from tensor maps was developed. Application of the method to the microscopic grayscale pulp images made it possible to detect curves corresponding to fibers in the pulp image and to compute their morphological characteristics. Performance of the method was evaluated based on the manually produced ground truth data. An accuracy of fiber characteristics estimation, including length, width, and curvature, for the acacia pulp images was found to be 84, 85, and 60% correspondingly.

Evaluation of Bubble Formation and Break Up in Suppression Pools by Using Pattern Recognition Methods

During a possible loss of coolant accident in BWRs, a large amount of steam will be released from the reactor pressure vessel to the suppression pool. Steam will be condensed into the suppression pool causing dynamic and structural loads to the pool. The formation and break up of bubbles can be measured by visual observation using a suitable pattern recognition algorithm. The aim of this study was to improve the preliminary pattern recognition algorithm, developed by Vesa Tanskanen in his doctoral dissertation, by using MATLAB. Video material from the PPOOLEX test facility, recorded during thermal stratification and mixing experiments, was used as a reference in the development of the algorithm. The developed algorithm consists of two parts: the pattern recognition of the bubbles and the analysis of recognized bubble images. The bubble recognition works well, but some errors will appear due to the complex structure of the pool. The results of the image analysis were reasonable. The volume and the surface area of the bubbles were not evaluated. Chugging frequencies calculated by using FFT fitted well into the results of oscillation frequencies measured in the experiments. The pattern recognition algorithm works in the conditions it is designed for. If the measurement configuration will be changed, some modifications have to be done. Numerous improvements are proposed for the future 3D equipment.

Magnetic Resocance Imaging Methods in the Follow-up of Multiple Sclerosis and in Farby Disease

Multiple sclerosis (MS) is a chronic immune-mediated inflammatory disorder of the central nervous system. MS is the most common disabling central nervous system (CNS) disease of young adults in the Western world. In Finland, the prevalence of MS ranges between 1/1000 and 2/1000 in different areas. Fabry disease (FD) is a rare hereditary metabolic disease due to mutation in a single gene coding α-galactosidase A (alpha-gal A) enzyme. It leads to multi-organ pathology, including cerebrovascular disease. Currently there are 44 patients with diagnosed FD in Finland. Magnetic resonance imaging (MRI) is commonly used in the diagnostics and follow-up of these diseases. The disease activity can be demonstrated by occurrence of new or Gadolinium (Gd)-enhancing lesions in routine studies. Diffusion-weighted imaging (DWI) and diffusion tensor imaging (DTI) are advanced MR sequences which can reveal pathologies in brain regions which appear normal on conventional MR images in several CNS diseases. The main focus in this study was to reveal whether whole brain apparent diffusion coefficient (ADC) analysis can be used to demonstrate MS disease activity. MS patients were investigated before and after delivery and before and after initiation of diseasemodifying treatment (DMT). In FD, DTI was used to reveal possible microstructural alterations at early timepoints when excessive signs of cerebrovascular disease are not yet visible in conventional MR sequences. Our clinical and MRI findings at 1.5T indicated that post-partum activation of the disease is an early and common phenomenon amongst mothers with MS. MRI seems to be a more sensitive method for assessing MS disease activity than the recording of relapses. However, whole brain ADC histogram analysis is of limited value in the follow-up of inflammatory conditions in a pregnancy-related setting because the pregnancy-related physiological effects on ADC overwhelm the alterations in ADC associated with MS pathology in brain tissue areas which appear normal on conventional MRI sequences. DTI reveals signs of microstructural damage in brain white matter of FD patients before excessive white matter lesion load can be observed on conventional MR scans. DTI could offer a valuable tool for monitoring the possible effects of enzyme replacement therapy in FD.

Multidata inverse problems and Bayesian solution methods in astronomy

Statistical analyses of measurements that can be described by statistical models are of essence in astronomy and in scientific inquiry in general. The sensitivity of such analyses, modelling approaches, and the consequent predictions, is sometimes highly dependent on the exact techniques applied, and improvements therein can result in significantly better understanding of the observed system of interest. Particularly, optimising the sensitivity of statistical techniques in detecting the faint signatures of low-mass planets orbiting the nearby stars is, together with improvements in instrumentation, essential in estimating the properties of the population of such planets, and in the race to detect Earth-analogs, i.e. planets that could support liquid water and, perhaps, life on their surfaces. We review the developments in Bayesian statistical techniques applicable to detections planets orbiting nearby stars and astronomical data analysis problems in general. We also discuss these techniques and demonstrate their usefulness by using various examples and detailed descriptions of the respective mathematics involved. We demonstrate the practical aspects of Bayesian statistical techniques by describing several algorithms and numerical techniques, as well as theoretical constructions, in the estimation of model parameters and in hypothesis testing. We also apply these algorithms to Doppler measurements of nearby stars to show how they can be used in practice to obtain as much information from the noisy data as possible. Bayesian statistical techniques are powerful tools in analysing and interpreting noisy data and should be preferred in practice whenever computational limitations are not too restrictive.

Cardiothoracic ratio and vertebral heart size (VHS) to standardize the heart size of the tufted capuchin (Cebus apella Linnaeus, 1758) in computerized radiographic images

Abstract: The VHS and CTR were assessed using computerized thoracic radiographs of ten clinically healthy tufted capuchin monkeys (five males and five females) from the Wild Animal Screening Center in São Luís (Centro de Triagem de Animais Silvestres de São Luís-MA-CETAS). Radiographs were taken in laterolateral and dorsoventral projections to calculate the cardiothoracic ratio (VHS) and vertebral heart size (CTR). The VHS showed mean values of 9.34±0.32v (males) and 9.16±0.34v (females) and there was no statistical difference between males and females (p>0.05). The CTR showed mean values of 0.55±0.04 (males) and 0.52±0.03 (females) and there was no statistical difference between the sexes (p>0.05). There was positive correlation between VHS and CTR (r=0.78). The thoracic and heart diameters showed mean values of 5.70±0.48cm and 2.16±0.40cm in the males, respectively. In the females they measured 5.32±0.39cm and 2.94±0.32cm. There was no statistical difference between the sexes. Our results show that the high correlation found between VHS and CTR permitted the verification with similar clinical precision between the two methods to estimate alterations in the heart silhouette by radiographic examination of tufted capuchin, making it an easy technique to apply that can be considered in the investigation of heart problems for this wild species.

The Shadow Moiré method using the phase shifting technique and digital image processing: computational implementation and application to the 3D-reconstruction of a buckled plate

The Shadow Moiré fringe patterns are level lines of equal depth generated by interference between a master grid and its shadow projected on the surface. In simplistic approach, the minimum error is about the order of the master grid pitch, that is, always larger than 0,1 mm, resulting in an experimental technique of low precision. The use of a phase shift increases the accuracy of the Shadow Moiré technique. The current work uses the phase shifting method to determine the surfaces three-dimensional shape using isothamic fringe patterns and digital image processing. The current study presents the method and applies it to images obtained by simulation for error evaluation, as well as to a buckled plate, obtaining excellent results. The method hands itself particularly useful to decrease the errors in the interpretation of the Moiré fringes that can adversely affect the calculations of displacements in pieces containing many concave and convex regions in relatively small areas.